An occupational health research group conducts a prospective cohort study to investigate t… — USMLE Step 3

An occupational health research group conducts a prospective cohort study to investigate the association between chronic exposure to a newly introduced synthetic resin and the development of interstitial lung disease (ILD). The study enrolls $4,500$ manufacturing workers, categorized into high-exposure and low-exposure groups based on their job roles, and follows them for $15$ years. At the conclusion of the observation period, the incidence of ILD is slightly higher in the high-exposure group. The calculated relative risk (RR) is $2.1$ , with a $95\%$ confidence interval (CI) of $0.8$ to $5.5$ and a $p$ -value of $0.12$ . The researchers publish a report concluding that the synthetic resin does not increase the risk of ILD.

Which of the following statistical concepts best explains the primary limitation of the researchers' conclusion?

Insufficient statistical power due to the low absolute number of observed disease events increases the probability of a Type II error.Answer
B
The confidence interval crossing $1.0$ indicates an elevated probability of a Type I error, which justifies the conclusion.
C
The wide confidence interval is a direct result of effect modification by unmeasured variables, which masks the true association.
D
The extended follow-up period artificially decreased the disease prevalence, which inversely lowered the positive predictive value of the exposure.
E
The study's reliance on observational data represents an anchoring bias that invalidates the statistical calculations.

Answer

Insufficient statistical power due to the low absolute number of observed disease events increases the probability of a Type II error.

The correct answer identifies that the study is likely underpowered. Statistical power is the probability of correctly rejecting the null hypothesis when a true difference exists (i.e., avoiding a Type II error). In a cohort study, power depends on the alpha level, the magnitude of the effect, and the sample size. Crucially, for rare outcomes like interstitial lung disease (ILD), power is primarily determined by the absolute number of incident events rather than the total number of enrolled subjects. Despite enrolling

4,500

workers, the rarity of ILD likely resulted in very few actual cases. This low event rate leads to high variance, generating a very wide confidence interval (

0.8

5.5

) and a non-significant

p

-value (

0.12

). A common statistical pitfall is concluding that 'absence of evidence is evidence of absence.' The authors' conclusion that the resin does not increase ILD risk is flawed because the wide CI and low power mean a clinically significant risk (e.g., RR of

2.1

) cannot be ruled out. Thus, the risk of a Type II error (a false negative) is high.

Step-by-Step Solution

Analyze the study results and the researchers' conclusion.

The study found an RR of

2.1

, but the CI crosses

1.0

(

0.8

5.5

) and

p > 0.05

. The researchers definitively concluded there is no risk based on this.

To identify the mismatch between the statistical findings and the authors' definitive conclusion.

Evaluate the relationship between outcome frequency, confidence intervals, and power.

ILD is a rare disease. In cohort studies with rare outcomes, power is limited by the absolute number of incident disease events, not just total enrollment. Few events cause wide CIs.

To determine why the study failed to find statistical significance despite a large relative risk point estimate.

Correlate the findings with standard statistical errors.

Failing to detect a true difference (a false negative) is a Type II error. The low power due to the rare outcome increases the risk of this error.

To identify the primary statistical limitation of concluding 'no risk' from a non-significant result.

Key Concept

In clinical studies, especially those evaluating rare outcomes, failure to achieve statistical significance (

p > 0.05

) may be due to low statistical power (Type II error) rather than a true absence of association. Statistical power in these scenarios is highly dependent on the absolute number of events.

Question

Answer

Step-by-Step Solution

Key Concept

Common Mistakes

Topics

More questions from this topic